Discussion

8.3 Results

significant difference between slow and fast RT for all of the five electrodes. All other pair-wise comparisons were only significant for ≤ 1 electrodes.

The previous analysis was repeated, but this time, RT groups were defined by shuffling RT values and by dividing these into thirds to serve as control. Using shuffled RT groups did not result in significant differences between any groups.

The analysis of the attentional modulation of EFPs from V1 revealed that the attentional modulation was higher in long lasting trials in comparison to shorter trials. In order to investigate whether the trial duration has an influence onto the EFP response latency, the EFP latency analysis was repeated with trials sorted by the time between cue onset and target LCA dimming (groups: short, medium and long trials). This analysis revealed a significant latency difference between the medium and the short trials group for only two electrodes (p < 0.025) and between the fast and the short trials group for a single electrode (p = 0.04).

8.4 Discussion

difficult tasks [Chen et al. 2008], only a low AI was found in the present study.

Despite the moderate hit rate, the RT median difference between the response in normal and catch trials was low. This could indicate two issues. Firstly, the task could have been too easy making a covert shift of attention unnecessary for the monkey.

The moderate hit rate of the monkey, however, indicates that the task was demanding.

Secondly, the low AI could have been caused by the monkey doing the task in a different than the desired way. The monkey could have found other strategies like broadening the attentional focus to cover large parts of the visual field. This behavior could be caused by the difficulty of the task. To encompass this issue, the number of objects needs to be reduced or the salience of the target change needs to be increased.

In the event of a too easy task, however, the monkey does not need to shift its attentional focus. In addition, a lower task demand in visual tasks increases the extent of the attentional focus [Handy et al. 1996]. A careful adaption of the task is necessary for future experiments. Higher RT differences between the target and the distractor response are preferable to be achieved in further investigations to ensure that the monkey does the task as desired.

2. Attentional effects are usually small in V1 compared to higher visual areas like V2 and V4 when investigated with intracortical electrodes (e.g., [Buffalo et al. 2010;

Luck et al. 1997; Mehta et al. 2000]). In addition, the signal amplitude of high-density ECoG was found to be higher in comparison to high-density EFPs [Bundy et al.

2014]. Therefore, the dura mater attenuates the signal recorded by high-density epidural electrodes which in turn leads to smaller signal differences between attend-in and attend-out, and thereby to a lower AI. To cope with the attenuation, other visual areas showing higher AI like V4 could be targeted. EFPs from V1 but predominantly V4 have been used for the decoding of the covert spatial attentional focus previously [Rotermund et al. 2013]. Keeping the task unchanged and recording from cortical areas that show a stronger attentional modulation in comparison to V1, could therefore enable a successful decoding of multiple spatial locations of the attentional focus from EFPs.

3. Some studies which found an attentional effect in V1 used visual stimulation which elicited γ-activity over a prolonged period of time (e.g., [Rotermund et al. 2009,

8.4 Discussion

2013]). A sustained γ-activity was absent in the present study due to the task design.

The task evoked predominantly transient γ-responses. Using an alternative visual stimulation that evokes a sustained neuronal activity could enhance the attentional effects in the EFP recorded from V1.

So far, the present attentional modulation is found in only one monkey. To verify the findings, this study needs to be repeated with additional monkeys.

A higher AI was found for trials in which the monkey scored faster RT and for trials with a longer time period between cue onset and target dimming. These findings are in line with a previous intracortical study in V1 showing a higher AI for longer delays of the target change [Sharma et al. 2015]. In addition, attentional modulation was shown to be higher in the visual area MT for trials in which the subject responded fast [Galashan et al. 2013].

The BGP activity was usually significantly modulated for electrodes belonging to the near and far RF eccentricity groups. The highest BGP activity in the attend-in condition was found for electrodes nearest to the target letter and BGP activity was decreasing with the distance from the target letter. Electrodes with an intermediate RF eccentricity did not show significant differences between the attend-in and -out condition. The AI was decreasing from positive values to negative values from near to far RF eccentricities analyzing all trials (Fig.

22). For all repetitions of the analysis, the negative AI of the most distant RF eccentricity (12°) increased to a less negative or even positive value. When ignoring the last finding, the attentional modulation of the γ-activity over eccentricities would support the gradient attention model [Mangun & Hillyard 1988; Shulman et al. 1986], which describes the modulation of spatial attention as a smooth decreasing function from the focus of attention to the surrounding area. The change in the sign of the AI comparing near and far RF eccentricities could be explained by the arrangement of the visual stimulation. The median distance between target and distractor was 9.8°, in a range of 5.6° to 12° (mean 8.7°).

Therefore, some far located electrodes from a target letter are potentially near to the distactor letter. Due to the fact that the attend-out condition for a target letter was an allocation of spatial attention to the distractor letter, the EFP of electrodes far from the target letter could have been modulated by the attentional focus onto the distractor. Furthermore, the less negative or positive AI for 12° RF eccentricity indicates less attentional modulation in comparison to groups with a lower RF eccentricity. This finding could be explained by the

8.4 Discussion

fact that only one out of 15 target had a distance of 12° to the distractor letter. For the other targets, which were more closely located to their distractor, the majority of electrodes which have an RF distance of 12° to the target letter are potentially not located near to the distractor.

In addition, only two target letters possessed more than nine electrodes with an RF eccentricity of 12° (see Fig. 21). Therefore, the reduced number of electrodes covering the spatial location of the attentional focus onto the distractor letters as well as the distance between targets and distractors may have caused the weaker negative or positive attentional modulation for the 12° RF eccentricity group.

The current analysis did not focus on the evoked transient amplitude in the EFP between normal and catch trials. In an EEG study where humans have been forced to attend a stimulus, the VEP amplitude increased [Groves & Eason 1969]. In a subsequent analysis, the transient response amplitude could be compared between EFP responses to targets in normal trials and the EFP response in catch trials where the catch was ignored. Furthermore, the comparison of the transient response amplitude between the target response in normal trials and the catch response in catch trials, provided that the catch and the target are the same letter, could reveal differences in the EFP response amplitude due to covert spatial attention.

In order to study the differences between EFPs evoked in normal and catch trials, a large amount of catch trials is necessary for each target letter, which was not provided in the present data set.

In the present study, a shorter latency of EFP onset responses evoked by the LCA dimming was found for trials with faster RT. The significant effect only occurred at five out of ten electrodes. Nevertheless, this result shows that EFP onset transients and behavior correlate.

The relationship between onset transients and RT was previously investigated using EEG and intracortical electrodes. The latency of the stimulus-induced VEP recorded with scalp electrodes from the occipital cortex was found to be shorter with increasing RT [Donchin &

Lindsley 1966]. In line with the previous finding, an EEG study found that different peaks of the ERP show shorter latencies when the RT was faster [Kammer et al. 1999]. They conclude that the latency modulation occurs at many stages of visual processing which is conclusive considering that different peaks of the EFP of EEG reflect different cortical stages of signal processing (for a review, see [Woodman 2010]). Comparing VEPs, RT and stimulus intensity, Vaughan and colleagues [1966] concluded that the velocity of RTs is defined by retinal and

8.4 Discussion

cortical processes, whereas cortical processes are represented by the VEP. Therefore, it is not surprising that another study showed that RT and VEPs both reflect the retinocortical processing time [Chakor et al. 2005]. Besides these interactions between RT and VEPs latency found in EEG, a few intracortical studies have investigated the relationship between onset latencies and RT. Amplitudes of pop-out responses were found to correlate to the behavioral performance, potentially reflecting the salience of the stimulus [Lee et al. 2002].

In addition, the amplitude of visually evoked responses in area MT was shown to correlate with RT [Galashan et al. 2013; Parto Dezfouli et al. 2018]. Due to the fact that RT and EFP onset latencies correlate, the present results show that the relationship between evoked transients and RTs found at the level of EEG and single cells in the visual cortex can also be concluded from EFPs recorded from V1.

Altogether, these findings suggest that attention and behavior can be correlated to the signal modulation of the EFP recorded from V1. This fact supports the findings of the previous chapters that emphasize the high information content which is represented in epidural signals.

Further investigations are necessary to investigate whether these modulations can be accurately and reliably decoded for single targets and single trials. A reliable encoding of the attentional focus from epidural signals recorded from V1 could pave the way for the development of meso-invasive gaze-independent BCIs.

9.Emphasizing the “positive” in positive reinforcement: using nonbinary rewarding for